WARNING: Tube/Valve amplifiers use potentially LETHAL HIGH VOLTAGES.
Building, troubleshooting and testing of these amplifiers should only be
performed by someone who is thoroughly familiar with
the safety precautions around high voltages.
In my experience I find just about any junk on the filament supply has an audiable impact , heaters on the other hand are another matter altogether . I have often thought about using switchers , 7N7 reminded me of their usefulness a while ago **after** I'd built a huge 5V @ 10.5A fialment supply for Eimac VT127A . Those cheap low voltage lighting 'transformers' look like a good place to start . I'd use these with a hefty laminated choke to provide some isolation from the supply . I'm sure this would make an excellent supply for firing up 845/211/713 etc
There are a few potential paths with switching filament supplies. One is to leave the AC square wave, and build a CT with a hum pot as is customary with 60 cps AC. Rectify the stuff comming from the wall through an isolation TX, wind a small step-down TX and switch the primary at 150kHz through a SE MOSFET. The gate-driver chips look quite easy to use. You want open-loop 50% duty cycle. The small, HF transformer will do a good job of isolating the filament from everything...no need for CMC as is often suggested for DC supplies.
Been and gone , torn down months ago . Tried VT127A zero bias at 600V with local cathode feedback . 7236 choke loaded CF driver and 809 input stage . Sounded nice but the output transformers ran on the ragged edge , really need to run 100mA 650V instead of 75mA at 600V . LCL filament supply weighed around 80lbs for both channels , so this project would never make it into a chassis
I've used switchers for both DHT filament heating and indirect heated types without issues. I do recommend some additional filtering on the output however to really kill the ripple at the output. Basically this consists of a small common mode choke right on the output of the supply followed by a cap, a normal choke, and another cap - these caps should have good performance at high frequencies, some small film types should be fine. (Place larger ele types in parallel - but do so only if you have a scope and look at the behavior at the ripple frequency. Typical CM chokes will be a few hundred uH to a couple of mH, and the normal mode choke a couple of mH typically. You actually can often buy these if you hunt around. The parts are compact so you are not looking at a lot of size or weight, but for best results you will need a scope and some patience to iterate the circuit for best results.
You can get ridiculously low levels of noise out of a switcher this way, and if you need even better performance use a standard linear regulator (!choke on input!) to get you to the precise voltage you need at the filaments.
I have access to a bunch of scrapped/obsolete/not-for-
production Class-D amps that technically "work". I think
some of them could fake a decent sine wave at 50KHz
or higher. The actual switching frame rates are way up
I had my eye on abusing them to light a pair of 814's.
Needing 10V and 3.5 amps apiece.
I built a line stage for a car. Both B+ and filament were ran off a switching supply.
Sure, it works, but it is very difficult to get performance that nears a linear supply. It take a lot of effort to remove all ringing and overshoot from a switching supply. There is a lot of harmonic content in square waves that makes them much more difficult to filter/work with when compared to a single sine wave.
As mentioned, capacitive coupling between the cathode and heater is an issue at switching frequencies. If powering any stages where the cathode is not bypassed with a capacitor, that noise will show up on the cathode. And if it's a stage with gain, or followed by a stage with gain, that noise will be amplified. Sure the frequency is out of the audible range, but it can intermodulate with the audible range, which alters the sound. Also, you can end up with a dirty looking signal, which to me isn't acceptable even if the noise is outside of the audible range.
Most switching based devices are measured using filtering which is why they get similar S/N ratios as linear based equipment. Without this filter it is probably impossible(at least very difficult) to get a S/N ratio that rivals it's linear equivalent.
There are really only three reasons that would justify going the switching route.
1- automotive application, which is a must
2- desperate need to shrink the transformer + reduce weight.
3- The need for better efficiency in a regulated power supply.
If your worried about noise, that should not be an issue... Most converters I build have switching fequency output ripple lower than .1% ...and that is over 200kHz....
If you are really nervous about the typical SMPS then use a resonant mode converter.... These use quasi-sine wave not square waves and the EMI emmisions are mucho lower lower than a standard converter.... I typically design them at or above 500kHz....so these would be very clean for filaments....
There is a more common design guide you can get from MAXIM. In principle, AC power is rectified to AC to the primary side of switching transformer mainly for safety to isolate wall power to device (it is also asked by regulation rule like UL). Also for this reason, an optical couple switch is used to control the on-off of switch transistor (BJT, or MOSFET).
The most simple and cheap approach is to use old AC adapter for mobile. Most of them is designed with 5V output. You can easily changed to 6.3V (or 6V) output by adjusting resistor values without problem.
SPECIAL Careful using switching power supply work with direct header
celsius235 reminds us a very important thing on applying switching power supply on direct heater. Previous discussion is mainly for isolated heater since AC/DC on heater may not directly applied to the amplification circuit. However, it is totally another story when switching power supply goes to direct heater.
In principle, I may not suggest using switching power supply for direct heater as commented by celsius235. However, if you want to try, it could be OK with special care mainly on block switching frequencies (very high freq) get into the amplification circuit. Besides on adding appropriate filters, two things should be noted
1. Switching freq should be higher as possible, > 300KHz is suggested. If possible, > 500KHz should be better but it comes the MOSFET selection and reduce switch loss. Except you are professional here, otherwise get a matual power module is recommended.
2. Capacitor used for by-pass switching noise is different from the one used for audio. Special polymar cap like SP-Cap, or OS-con will be appropriate for their extremely high ripple current. However, most of them can only work at low voltage (but no de-rating issue).
I recently have the same thought. What about using switching power supply with voltage control and current control? Do you think this could help DIYer to build better DHT amps if the product is available at around US$25 per module?
There are already some high frequency switched mode power supply boards targeted at car audio tube amplifiers, but I only see them on Taobao, I newer saw them on sale on popular international ecommerce sites yet. See as example this one: 膽機 前級 電子管 DC12V升壓 逆變 電路板
Current regulation for filaments is not needed, altough a current limiter function may be useful to soft-start the filaments turn-on.
A benefit of most switched mode topology converters is the ease to add multiple outputs. Having both the filament and the HT power supplies from a small SMPS will remove the need for a power transformer and will reduce the chassis size. Several small SE amps sold from China already use a switched mode power supply, but I haven't found any with bigger tubes such as KT88 or even EL34. I would be really interested on a power supply board like this. At some point I was almost ready to design one myself when I had to use small chassis, but ultimately I reverted to a traditional linear supply and some creative placement of the transformers.
Right now not only am I using a 12V SMPS for the heaters of 20 tubes, the B+ is created from 12V using boost converters. Works perfectly. If you're worried about HF hash making it into the audio path use a filter choke (50mH or so) 6550 PP outputs.